1999 年 65 巻 634 号 p. 2561-2567
In automatic inspection of X-ray radiographic testing, it is difficult to detect small and unclear defect images from low contrast film. Generally, these defect images are hidden by noise due to unclear background image, small particles of film, uneven development of film and so On. Furthermore, it is difficult to abstract shape of defects, because the boundaries among defect images and their background images are unclear. Therefore, in order to eliminate background image and detect defect boundaries surely, two dimensional background subtraction method and region growing method are proposed. This paper deals mainly with the algorithms of the background subtraction method and the region growing method, and their effectiveness of these methods in automatic inspection of X-ray films.